王峰 AI × 经营 约一次诊断

总目 / 文章 / WF-0385

文章 · 存档

不确定性,是人类在AI面前最后一尺遮羞布

这句话有点自嘲,又有点倔强。好像我们站在一台无比强大的机器面前,身上的铠甲一块块被剥掉。什么记账速度、报表准确率、预测模型、流程优化,全都在AI的计算力和算法面前变得苍白。最后只剩下一块薄薄的布,还在死撑着体面。这块布,叫“不确定性”。

如果把过去一百年的管理史摊开看,你会发现人类一直相信“确定性”才是专业的象征:标准化流程、精细化管控、可以复制的经验、可预测的回报。

财务人靠会做账、会报表、会预算吃饭,经营者靠会做计划、会控制成本、会算投入产出立身。那时候谁能把事情做到心中有数,谁就更像一个靠得住的人。

但AI悄无声息地把这套逻辑翻了个面。所有“有数”的东西都开始被机器接管。

凭证自动生成,报表一键汇总,预算模型自动跑模拟,现金流压力自动预警,甚至连税务筹划、资金配置、风控场景,都有越来越成熟的算法在跑。

那些曾经让人类觉得自己挺厉害的手艺,正在被系统和模型一点点吞掉。

你有没有想过一个问题,只有不确定性的存在才需要人类决策,确定性的老早都被自动化完整取代了。

而现存的不确定性正在逐渐减少。还好,幸运的是随着AI不断放大知识和能力认知的边界,越来越多的不确定性如雨后春笋般冒出来。

确定性的部分完全不需要人;还没被吃干净的确定性也只是时间问题。

那些可以靠规则、算式、流程解决的问题本质上都属于静态题库,AI只是在毫不费力地不断刷题。

而只有答不完、算不清、没有标准答案的那一部分才真正轮到人类出场。

你看现在一线企业里的AI业财实践,真正拉开差距的地方从来不是谁报表做得更快,而是谁敢在AI给出的信息基础上做出那个没人替你兜底的选择。

AI可以在三秒钟里帮你算出十种情景下的利润、现金流和资产负债表压力,可以告诉你不同定价策略对客户分层、订单结构和毛利率的影响,可以把行业数据、宏观数据、公司历史数据一股脑儿摆在你面前,但它永远不会替你签字那一笔:到底要不要关掉某个大客户的低毛利业务,要不要在明年冬天押注一个看上去不那么安全的新市场,要不要为了研发压缩当期利润、赌三年后的技术曲线。

这就是不确定性的价值所在。

它不是一个学术词,甚至不是一个抽象概念,而是一种非常具体的感受:你知道AI已经把能算的都算了,把看得见的坑都标出来了,但你心里仍然有一团东西,叫谁也保证不了。那一团东西就是人的战场。

更有意思的是,AI并没有像很多人恐惧的那样把未知压扁,反而是随着AI不断放大知识和能力认知的边界,越来越多的不确定性如雨后春笋般冒出来。

原来我们以为我们不知道的是“我们不知道什么”;现在AI帮我们照亮了很多角落,我们才发现,原来还有那么多没被问过的问题、没被试过的路径、没被看见的可能性。

以前一家公司做决策,信息是严重不对称的,决策者更多是在浓雾里凭经验摸索,所谓拍板,其实很多时候是一种无奈的赌。

现在不同了,数据湖、实时BI、AI分析助手、智能预测模型,这些东西把信息缺失这个借口一点点夺走。

你不能再轻易说当初不知道,因为大量信息明明摆在那儿。可是,即便如此,真正关键的那几步,依然没人能替你走。

回到我自己的专业领域,AI业财实践越成熟这个矛盾也就反而是越清晰:信息越来越全,模型越来越准,场景越来越细,风险提示越来越前置,但决定要不要压下那一笔投资、要不要削掉那块看起来还行的业务、要不要重构一条已经跑顺的供应链,这些事仍然不会有一个绝对正确的答案。AI做得再漂亮,最后仍然要有人出来说:就这样,干。

这就是人类在AI面前最后一尺遮羞布的真实含义。

不是我们还掌握着什么别人拿不到的技术,而是我们还必须承担那部分没有技术指标、只有后果的东西。

遮羞布遮住的是人类永远都难以超越一个AI对这个世界上所有信息的整合能力,遮羞布遮住的甚至只是人类大脑永远都无法超越一块硬盘存储信息的数量。

但这个遮羞布遮不住的,也恰恰就是人类之所以被称之为人类的独立思考与创新创造,以及对未知事务的决断力和责任。是勇气,是价值观,是“你到底愿意为哪种结果负责”。

从某种意义上讲,AI把财务和管理人员从算账的人逼成了赌未来的人。

以前你可以躲在Excel算力有限的后面,躲在会计准则复杂难懂的后面,躲在公司流程错综冗长的后面,现在这些东西都自动化了,反而把你推到台前。你再也没法只做一个执行者,而是不得不变成那个对不确定性有判断、有坚持的人。

这也是为什么真正走在前面的企业,对AI业财的要求绝不是上几套工具、建几个模型、搞几张酷炫的大屏,而是逼着财务业务和管理层一起把那些最难最不确定最没人敢负责的问题摊开来谈:

如果AI告诉你现有模式三年后极有可能利润下滑,你敢不敢现在动刀?

如果AI模拟出一个更激进但更有增长空间的战略路径,你敢不敢背着短期利润的压力试一次?

如果AI把一些你习惯依赖的老经验推翻了,你承不承认自己过去可能看错了?

不确定性从来不是好看的,它很丑,它让人难受,让人纠结,让人夜里睡不着觉。

但也正因为它难看,它才值得被人类拿在手里。

AI可以成为最强大的放大镜,把所有信息、所有可能性、所有路径摊开在你眼前,可是谁伸手去抓、抓哪一条路,那一刻不是算法,而是人。

真正值得期待的未来,大概会是这样一种画面:

确定性的部分被AI做得干干净净、极致高效,不再消耗人类的时间;

而不确定性的部分被人类当成真正的战场,学会用AI把风险看得更清楚,把机会看得更早,把决策做得更有底气,但从不把责任推给机器。

人类不是输给AI,而是被AI逼着终于不得不长大。

如果有一天,连不确定性都不需要人类了,那才是真正可怕的世界。

现在至少我们还可以大大方方承认:是的,我需要这最后一尺遮羞布,因为我还需要一点保留,一点犹豫,一点只属于人的心跳。

但只要这块布还在我们手里,AI再强,也只是工具,而永远都不会是结局的主人。

Uncertainty is the last little cloth

Uncertainty is the last little cloth that humans use to cover ourselves in front of AI

This sentence is a little like making fun of ourselves, and also a little stubborn. It is like we stand in front of a super strong machine, and our armor is taken off one piece at a time. Things like fast accounting, perfect reports, prediction models, and better work steps all look weak in front of AI's big math and strong computer power. In the end, we only have a very thin cloth left to hold our pride. This cloth is called “uncertainty.”

If we look at the last one hundred years of management history, we can see that people always think “certainty” means someone is a real expert: standard steps, careful control, experience you can copy, and results you can guess before they come.

People in finance make money because they can do the books, make reports, and plan budgets. Business leaders make money because they can plan, control cost, and count money going in and out. In the old days, the one who knew everything and could plan well looked like the person you could trust most.

But AI quietly changed this idea. All the things that can be counted are now done by machines.

Proofs are made by the system, reports come with one click, budget models run by themselves, cash problems show alerts by themselves, and even tax planning, money use, and risk checks are done by better and better computer models.

The skills that once made humans feel proud are being eaten away by systems and models little by little.

Think about this: only things with uncertainty need humans to choose. Things that are 100% sure are already taken by machines.

But the uncertainty we still have now is becoming smaller. Luckily, when AI makes our knowledge and power bigger, more new uncertainty grows like new plants after a big rain.

Things that are fully sure do not need people at all; things that are not fully eaten by AI will be eaten later.

All problems that can be solved by rules, math, or steps are like a big test book. AI is just doing more and more questions easily.

Only the part that has no final answer and no perfect math can be done by humans.

Look at AI in top companies right now. The real big difference is not who makes reports faster, but who can make the choice that no machine will take the blame for, even after AI gives all the info.

AI can tell you profit, cash, and stress of the company in ten different cases in three seconds. It can tell you what happens if you change price: what customers you get, what orders change, and how much money you make. It can put industry data, big world data, and company history all in front of you. But it will never sign the final paper: should we close a big customer who has low profit? Should we try a new market next winter that does not look safe? Should we lose money now to make new tech that maybe works three years later?

This is where uncertainty has value.

It is not a school word, and not just an idea. It is a real feeling: you know AI already counted all the things it can count, and marked all the dangers it can see, but there is still something inside you that no one can promise. That thing is the human battlefield.

And the funny thing is, AI does not kill the unknown like many are scared of. When AI makes our knowledge bigger, more new uncertainty comes out.

Before, we thought the unknown was “we don't know what we don't know.” Now AI puts light on many dark corners, and we see many more questions we never asked, paths we never tried, and chances we never saw.

Long ago, when a company made choices, information was very unfair. Leaders made choices in deep fog using only experience, and “making the final call” was often just a helpless guess.

Now things changed. Data lakes, fast business info, AI helpers, and smart prediction models take away the excuse of “I did not know.”

You cannot say you did not know, because the information is right there. But even so, the most important steps still must be walked by a person.

Back to my own work field. The better the AI finance work becomes, the clearer this truth gets: information is more complete, models are more correct, plans are more detailed, and warnings come earlier, but whether to spend money on a new plan, cut a business that still works okay, or rebuild a supply chain that is already smooth — there is still no “perfect answer.” AI can do great work, but a human must still say: Yes. We go.

This is what it really means that uncertainty is the last little cloth humans use to cover ourselves in front of AI.

We do not win because we have some magic skill. We must still take the part where there are no numbers, only results.

The cloth covers that humans will never beat AI in knowing and saving all the information in the world. The cloth covers that our brain will never save more than a computer.

But the cloth cannot cover the one thing that makes humans human: the power to think new things, to create new ideas, to choose when we do not know the end, and to take the blame. It is courage, it is values, it is “What result are you willing to stand for?”

In some way, AI pushes finance and managers to become people who bet on the future.

Before, you could hide behind slow Excel, hard accounting rules, and long company steps. Now all these things are automatic, and AI pushes you in front. You must stop being just a worker. You must be the person who can think and decide when things are unsure.

This is why the best companies do not just want AI tools and cool screens. They push finance people, business people, and leaders to talk about the hardest problems with the biggest uncertainty:

If AI tells you the company will make less money in three years, do you dare fix the problem now?

If AI shows a risky plan that can grow big later, do you dare try even if you make less money this year?

If AI shows your old ideas are wrong, can you say you were wrong?

Uncertainty is never pretty. It feels bad. It makes you worry. It keeps you awake at night.

But because it feels bad, humans must hold it.

AI can show all info, all paths, and all choices, but who grabs the chance and which one they grab — that is not the computer. That is a person.

A really good future may look like this:

AI does every boring thing that is sure and does it super well, so people do not waste time anymore;

And people use the unsure part as their real battlefield, using AI to see danger earlier, see chances earlier, and decide with more confidence, but never tell the machine to take the blame.

Humans do not lose to AI. AI makes humans finally have to grow up.

If one day, even uncertainty does not need humans, that will be the scary world.

Right now, we can still say this with an open heart: Yes, I need this last little cloth, because I still need a little secret, a little fear, and a little heartbeat that only belongs to humans.

But as long as this cloth is still in our hands, no matter how strong AI is, it will always be a tool, and never the boss of the ending.